Sprouty2 Protein Regulates Hypoxia-inducible Factor-α (HIFα) Protein Levels and Transcription of HIFα-responsive Genes

Abstract

The α-subunits of hypoxia-inducible factors (HIF1α and HIF2α) promote transcription of genes that regulate glycolysis and cell survival and growth. Sprouty2 (Spry2) is a modulator of receptor tyrosine kinase signaling and inhibits cell proliferation by a number of different mechanisms. Because of the seemingly opposite actions of HIFα subunits and Spry2 on cellular processes, we investigated whether Spry2 regulates the levels of HIF1α and HIF2α proteins. In cell lines from different types of tumors in which the decreased protein levels of Spry2 have been associated with poor prognosis, silencing of Spry2 elevated HIF1α protein levels. Increases in HIF1α and HIF2α protein levels due to silencing of Spry2 also up-regulated HIFα target genes. Using HIF1α as a prototype, we show that Spry2 decreases HIF1α stability and enhances the ubiquitylation of HIF1α by a von Hippel-Lindau protein (pVHL)-dependent mechanism. Spry2 also exists in a complex with HIF1α. Because Spry2 can also associate with pVHL, using a mutant form of Spry2 (3P/3A-Spry2) that binds HIF1α, but not pVHL, we show that WT-Spry2, but not the 3P/3A-Spry2 decreases HIF1α protein levels. In accordance, expression of WT-Spry2, but not 3P/3A-Spry2 results in a decrease in HIF1α-sensitive glucose uptake. Together our data suggest that Spry2 acts as a scaffold to bring more pVHL/associated E3 ligase in proximity of HIF1α and increase its ubiquitylation and degradation. This represents a novel action for Spry2 in modulating biological processes regulated by HIFα subunits.

Introduction

The four Sprouty proteins (Spry1 to Spry4), which are products of different genes, regulate downstream signaling from certain receptor tyrosine kinases and therefore play a major role in development (1–5). Because cell proliferation processes in normal development and tumor growth overlap, some of the Spry² proteins, such as Spry1, Spry2, and Spry4, also regulate tumor growth (6–16). Among the four isoforms, Spry2 is ubiquitously expressed and well studied in cancer. In cancers of the liver, lung, breast, and prostate, Spry2 levels are decreased (6–11, 17–19). The decrease in Spry2 levels in these cancers has been correlated to poor patient prognosis and shorter survival of patients implicating Spry2 as a tumor suppressor. Therefore, a number of studies have investigated mechanisms that regulate the expression of Spry2, as well as how Spry2 modulates signaling via receptor tyrosine kinases (20–25). At the post-translational level, Spry2 has been shown to be ubiquitylated and targeted for proteosomal degradation by c-Cbl (26, 27), Siah2 (28), Nedd4-1 (29), and pVHL with its associated E3 ligase (30). Interestingly, in some patient-derived hepatocellular carcinomas, when Spry2 levels were decreased, the levels of Nedd4-1 were elevated (8).

It is now well established that as tumors proliferate rapidly prior to angiogenesis, the oxygen from the nearby vasculature cannot diffuse throughout the entire tumor resulting in the formation of a hypoxic environment (31). Cells adapt to the hypoxic environment by altering a variety of cellular processes that are heralded by an increase in the levels of the transcription factors, hypoxia-inducible factors (HIFs) (32–36). HIFs are heterodimers composed of one of three α-subunits (HIF1α, HIF2α, and HIF3α) and a β-subunit (HIF1β). The role of HIF3α in hypoxic gene regulation is not well understood, but a splice variant of HIF3α may function as an inhibitor of gene transcription during hypoxia (37, 38). In normoxia, HIF1α and HIF2α are degraded by a well characterized proteosomal mechanism (39, 40). Prolyl hydroxylase domain proteins (PHDs) bind to the HIFα subunits and hydroxylate two proline residues (41, 42). These hydroxy-prolyl residues serve as the docking site for pVHL and its associated E3 ubiquitin ligase resulting in the ubiquitylation and subsequent proteosomal degradation of HIFα subunits (43–46). Because the hydroxylation reaction requires oxygen, in hypoxia the activities of the PHDs are attenuated, and therefore, the hydroxylation, ubiquitylation, and degradation of HIFα subunits are inhibited. This permits the accumulation of HIFα protein levels and consequently enhances their ability to regulate transcription of genes, such as those that regulate proliferation, angiogenesis, drug metabolism, and glycolysis to promote tumor survival and growth (47–49).

Because of the anti-proliferative and tumor suppressor actions of Spry2, we investigated whether Spry2 alters the levels of HIFα subunits and/or their ability to alter transcription of their target genes. Herein, we report the novel ability of endogenous Spry2 to promote the degradation of HIF1α and HIF2α resulting in a decrease in the expression of the genes that the two HIFs regulate. With HIF1α as the prototype of the two HIFα proteins (HIF1α and HIF2α), we demonstrate that Spry2 decreases HIF1α protein stability and enhances the ubiquitylation of HIF1α. Using different approaches, we also show that Spry2 exists in a complex with HIF1α. Because we previously reported that Spry2 is capable of binding to pVHL (30), we hypothesized that Spry2, by binding with HIF1α, increases the amount of pVHL in the vicinity of HIF1α and promotes its degradation through a pVHL-dependent mechanism. Indeed, our studies show that when pVHL is silenced, Spry2 cannot alter HIF1α protein levels. Additionally, a mutant form of Spry2 that does not bind to pVHL also does not decrease HIF1α protein levels. In agreement, wild-type (WT) Spry2 inhibits HIF1α-sensitive glucose uptake, whereas the mutant form of Spry2, which does not associate with pVHL, does not alter glucose uptake.

Results

Endogenous Spry2 Regulates HIF1α and HIF2α Protein Levels

Because previous studies have suggested that Spry2 plays a critical role in the regulation of hepatocellular, prostate, breast, and lung carcinomas, we utilized HuH7 cells that are derived from a human hepatocellular carcinoma (6–11, 17–19). In these cells, the siRNA-mediated silencing of endogenous Spry2 resulted in a 2.5-fold elevation of endogenous HIF1α protein levels (Fig. 1A). As established by numerous studies, HIF1α levels are not detectable in normoxia. However, once the cells are exposed to hypoxia, HIF1α protein levels are elevated (Fig. 1A). It is also important to note that as reported in our earlier report (30), Spry2 levels are also elevated by hypoxia (Fig. 1A). The increase in HIF1α protein levels upon silencing of endogenous Spry2 is not accompanied by a change in HIF1α mRNA levels (Fig. 1B). Like HIF1α, endogenous HIF2α protein levels in HuH7 cells were also increased by hypoxia and further elevated by 2-fold upon silencing of Spry2 (Fig. 1C) without any changes in HIF2α mRNA levels (Fig. 1D). A second Spry2 siRNA targeting a different sequence in the mRNA also elevated HIF1α and HIF2α protein levels in HuH7 cells (Fig. 2A). Notably, Spry2 is modified by phosphorylation and migrates as multiple bands (50), which depending on length of exposure of the blots can be observed as a single band or a doublet (cf. Figs. 1, A and B, and 2A). Together, these findings suggest that Spry2 alters HIF1α and HIF2α protein levels post-transcriptionally.

FIGURE 1.

Silencing of Spry2 increases HIF1α and HIF2α protein amounts without altering their mRNA levels. HuH7 cells transfected with control (Cntrl) siRNA or siRNA targeting Spry2 were cultured in normoxia (N) (21% O₂) or hypoxia (H) (3% O₂) for 8 h (A and B) or 24 h (C and D). Representative Western blots probed for HIF1α (A), HIF2α (images from same blot) (C), Spry2, and actin (loading control) are shown. Graphs in lower panels show mean ± S.E. of densitometric analysis of HIF1α (A) or HIF2α (B) normalized to actin from three (A) or four (C) independent experiments. Transcript levels of HIF1α (B), HIF2α (D), and SPRY2 were monitored using qRT-PCR, and relative levels were calculated as described under “Experimental Procedures.” Graphs are mean ± S.E. from three independent experiments. Statistical significance was assessed using unpaired Student's t tests (A and C); n.s., not significant.

FIGURE 2.

Two siRNAs against Spry2 enhance HIF1α protein levels in a variety of cell lines, and ectopic expression of Spry2 decreases HIF1α protein levels. A, HuH7 cells transfected with control (Cntrl) siRNA or a different siRNA targeting Spry2 were cultured in hypoxia (H) (3% O₂). Western blots probed for HIF1α, HIF2α, Spry2, and actin (loading control) are shown. B, MCF7, MDA-MB-231, A549, and Hep3B cells transfected with control (−) or one of two different Spry2 siRNAs (1 and 2) were cultured in hypoxia (3% O₂). A representative blot probed for HIF1α, Spry2, and actin and captured with the Bio-Rad ChemiDoc XRS+ instrument is shown. Graph shows mean ± S.E. of densitometric analysis of HIF1α normalized to actin from three independent experiments. C, control or HA-Spry2-expressing HeLa cells were cultured in normoxia (N) (21% O₂) or hypoxia (H) (3% O₂). Representative blot probed for HIF1α, HA-Spry2, and ERK1/2 (loading) is shown. Graph below shows mean ± S.E. of densitometric analysis of HIF1α normalized to actin from four independent experiments. Statistical significance was assessed by unpaired Student's t test (B and C).

The generality of our findings is exemplified by the observations that silencing of Spry2 with two different siRNAs results in an elevation of HIF1α protein levels in the following panel of cell lines derived from tumors in which Spry2 plays a crucial role: breast cancer-derived cell lines (MCF-7 and MDA-MB-231); lung cancer cell line (A549); and another hepatocellular carcinoma cell line (Hep3B) (Fig. 2B). Different cell types express different amounts of Spry2 and different amounts of post-translationally modified Spry2. As mentioned above, Spry2 migrates as a doublet representing its phosphorylated and dephosphorylated forms (29, 50), and thus, it would appear that in A549 cells a larger amount of Spry2 is phosphorylated, whereas in Hep3B cells a larger amount of dephosphorylated Spry2 is present (Fig. 2B). Likewise, HIF1α is also phosphorylated and migrates as a doublet (51–53), and depending upon the cell type, the relative amounts of phosphorylated and dephosphorylated HIF1α as well as the total amount of HIF1α vary. In contrast to the findings with Spry2 silencing, the ectopic expression of Spry2 in HeLa cells, which do not express detectable protein levels of endogenous Spry2, decreased HIF1α protein levels (Fig. 2C). Moreover, expression of the other isoforms of Spry (Spry1, Spry3, and Spry4) also decreased HIF1α protein levels (data not shown). These findings suggest that all isoforms of Spry proteins regulate HIF1α protein levels. However, to elucidate the mechanisms and impact of the regulation of HIFα protein levels, herein we have focused on Spry2 as a prototypic member of this family.

Spry2 Regulates the Ability of HIF1α and HIF2α to Modulate the Transcription of Their Target Genes

HIF1α and HIF2α form heterodimers with HIF1β and act as transcription factors to regulate a variety of genes, including those that regulate glycolysis (54–60). To determine whether the ability of HIF1α and HIF2α to function as transcription factors was altered upon silencing of endogenous Spry2, mRNA levels of HIF1α- and HIF2α-regulated genes were monitored using quantitative RT-PCR. As examples of HIF1α-regulated genes, we monitored the mRNA levels of the glucose transporter GLUT1 and the glycolytic enzyme phosphoglycerate kinase 1 (PGK1). Consistent with the changes in protein levels of HIF1α, Spry2 silencing resulted in a significant elevation of GLUT1 and PGK1 mRNA levels (Fig. 3, A and B). The silencing of HIF1α, but not HIF2α, abolished the ability of siRNA against Spry2 to elevate GLUT1 and PGK1 mRNA levels demonstrating that silencing of Spry2 mediated its effects on GLUT1 and PGK1 via HIF1α and not HIF2α (Fig. 3, A and B).

FIGURE 3.

Spry2 silencing enhances the transcript levels of HIF1α- and HIF2α-regulated genes and activity of the PFK-Luc reporter. HuH7 cells were transfected with control siRNA or siRNAs against Spry2, Spry2 + HIF1α, or Spry2 + HIF2α and cultured in hypoxia (3% O₂). Transcript levels of GLUT-1 (A), PGK1 (B), EPO (D), SPRY2 (E), HIF1α (F), and HIF2α (G) were monitored using qRT-PCR, and relative levels were calculated as described under “Experimental Procedures.” C, HuH7 cells were transfected with the same siRNAs as described for A, B, and D along with pGL2-Pfkfb3/−3566 and pRG-TK (as control) and cultured in normoxia (21% O₂) or hypoxia (3% O₂) for 24 h. Luciferase assays were performed as described under “Experimental Procedures.” Data shown are the mean ± S.E. from four (B and D–G), three (A), or at least three (C) independent experiments. Statistical significance was assessed by one-way ANOVA with Tukey's multiple comparisons (A, B, and D–G) or two-way ANOVA with Sidak's multiple comparisons (C).

As a second approach to monitor the ability of Spry2 to modulate HIF1α-elicited transcription, we transfected HuH7 cells with a luciferase reporter plasmid composed of the phosphofructose kinase (PFK) promoter fused to a luciferase reporter gene (PFK-Luc) and a Renilla luciferase plasmid as a control for transfection efficiency. Transcription of the PFK gene and the PFK promoter-Luc construct is regulated by HIF1α but not by HIF2α (58). In these cells, the ability of HIF1α to regulate transcription, as measured by luciferase activity, was elevated 2.5-fold by hypoxia (Fig. 3C). Moreover, silencing of Spry2 further elevated luciferase activity by 63% (Fig. 3C). Silencing of HIF1α, but not HIF2α, abolished the ability of Spry2 targeting siRNA to enhance luciferase activity (Fig. 3C) demonstrating that silencing of Spry2 enhances the transcription of PFK-Luc via changes in HIF1α levels. These data (Fig. 3, A–C) also show that, as reported by others (54, 55, 58, 59), the expressions of GLUT1, PGK1, and PFK-Luc are regulated by HIF1α and not HIF2α.

Recently, erythropoietin (EPO) was identified as an HIF2α-responsive gene (61, 62). Therefore, to determine whether the enhanced HIF2α levels observed upon silencing of Spry2 (Fig. 1B) also altered the transcription of the HIF2α-responsive EPO gene, we monitored EPO mRNA levels in HuH7 cells. Akin to the findings with GLUT1 and PGK1 for HIF1α, the silencing of Spry2 elevated EPO mRNA levels, and the silencing of HIF2α, but not HIF1α, obliterated the ability of the siRNA against Spry2 to elevate EPO mRNA levels (Fig. 3D). The efficient silencing of the SPRY2, HIF1α, and HIF2α expression by the different siRNAs for the qRT-PCR data is shown in Fig. 3, E–G. Overall, the data presented thus far demonstrate that Spry2 decreases HIF1α and HIF2α protein amounts without altering the mRNA levels of HIF1α or HIF2α, and when Spry2 expression is silenced, the elevation in their protein levels increases the transcription of their respective target genes. Hence, endogenous Spry2 regulates the endogenous HIFα protein content and the transcription of their target genes.

Spry2 Regulates the Stability and Ubiquitylation of HIF1α

The lack of any changes in mRNA levels of HIF1α or HIF2α when expression of Spry2 is silenced in the face of a 2–2.5-fold increase in the levels of the two proteins (Fig. 1) would suggest that Spry2 regulates HIF1α and HIF2α post-translationally. The post-translational regulation of HIF1α and HIF2α involves very similar mechanisms consisting of hydroxylation of two prolyl residues that serve as the binding site for von Hippel-Lindau protein (pVHL) and associated E3 ligase (39, 40). Therefore, using HIF1α as a prototypic member, we first examined whether Spry2 altered the stability of the HIF1α protein. Essentially, HuH7 cells transfected with control siRNA or siRNA against Spry2 were treated with the protein translation inhibitor cycloheximide, and the levels of HIF1α were monitored at different times. As shown in Fig. 4A, silencing of Spry2 increased the half-life of HIF1α from 4.3 to 7.1 min. The major degradation pathway for HIF1α involves ubiquitylation followed by proteosomal degradation (63). Therefore, we directly determined whether Spry2 altered the ubiquitylation of HIF1α. For this purpose, we utilized HEK293T cells because the transfection efficiency of HuH7 cells with FLAG-ubiquitin was very low. HEK293T cells were transfected with FLAG-ubiquitin and either control siRNA or siRNA against Spry2. The whole cell lysate (WCL) shows efficient silencing of Spry2 and equal expression of FLAG-ubiquitin (Fig. 4B, left panel). As mentioned earlier, HIF1α appears as a single or double band depending upon its phosphorylation state (see Fig. 2B). In HEK293T, cells we only observe a single band; whether this is the phosphorylated or dephosphorylated HIF1α is difficult to determine. Immunoprecipitation of HEK293T cell lysates under denaturing conditions with nonspecific IgG or anti-FLAG antibody showed that the anti-FLAG antibody immunoprecipitates similar amounts of total FLAG-tagged ubiquitylated proteins from cells treated with control siRNA and siRNA against Spry2 (Fig. 4B, right panel). Probing the immunoprecipitates for HIF1α assessed the amount of ubiquitylated HIF1α in each sample. When Spry2 was silenced, the amount of ubiquitylated HIF1α was ∼50% lower than that in control siRNA-transfected cells (Fig. 4B, right panel and bar graph at bottom). These data (Fig. 4) suggest that endogenous Spry2 decreases HIF1α stability by increasing the ubiquitylation of HIF1α and targeting it for proteosomal degradation.

FIGURE 4.

Spry2 decreases the stability of HIF1α by enhancing the ubiquitylation of HIF1α. A, HuH7 cells transfected with control (Cntrl) siRNA or siRNA targeting Spry2 were cultured in hypoxia (3% O₂) for 8 h. Cycloheximide (CHX) (200 μm) was added, and cells were lysed at the indicated times. Representative Western blots probed for HIF1α, Spry2, actin, and ERK1/2 are shown. Graph shows mean ± S.E. of densitometric analysis of HIF1α normalized to the average of actin and ERK1/2 from five independent experiments. Half-lives are listed in legend. Statistical significance was assessed by unpaired Student's t test. B, HEK293T cells were transfected with constructs of HIF1α and FLAG-tagged ubiquitin (Ub) along with either control (Cntrl) or Spry2 siRNAs. Cells were cultured in hypoxia (3% O₂) for 16 h, and MG132 (25 μm) was added in during the last 4 h of incubation. All FLAG-tagged ubiquitylated proteins were IP, and the amount of ubiquitylated HIF1α was monitored by Western blotting analyses; immunoprecipitation with nonspecific mouse IgG served as control. WCL were analyzed for HIF1α, Spry2, FLAG-tagged ubiquitin, and actin (loading). Graph shows mean ± S.E. of densitometric analysis of ubiquitylated HIF1α divided by total immunoprecipitated FLAG-tagged ubiquitylated proteins from three independent experiments. Statistical significance was assessed by unpaired Student's t test.

Spry2 Exists in a Complex with HIF1α and Regulates HIF1α through pVHL-dependent Mechanism

To identify the mechanism by which Spry2 regulates HIF1α ubiquitylation and degradation, we determined whether endogenous Spry2 and HIF1α exist in the same complex. In hypoxic HuH7 cells treated with the proteosomal inhibitor MG132, immunoprecipitates of Spry2 contained endogenous HIF1α (Fig. 5A). Additionally, by immunocytochemistry, we also observed the co-localization of Spry2 and HIF1α, and this co-localization was markedly decreased upon silencing of either HIF1α or Spry2 (Fig. 5B). The co-localization of Spry2 and HIF1α within intact cells was further confirmed using the proximity ligation assay (PLA) that permits detection of interacting proteins within intact cells (64). As shown in Fig. 5C, the PLA approach also showed that Spry2 and HIF1α are in close proximity, and the PLA signal is markedly diminished when HIF1α is silenced. Interestingly, the data in Fig. 5, B and C, show that Spry2 and HIF1α interact in the nucleus. Although the localization of Spry2 in cytoplasm and membrane ruffles has been previously reported (65, 66), this is the first demonstration of Spry2 being present in the nucleus. Overall, using three different approaches, the data in Fig. 5 demonstrate that endogenous Spry2 and HIF1α exist in a complex.

FIGURE 5.

Spry2 exists in a complex with HIF1α. A, HuH7 cells incubated in hypoxia (3% O₂) for 8 h were treated with MG132 (25 μm) during the last 4 h of the incubation. Spry2 was IP, and the co-immunoprecipitation of HIF1α was monitored; nonspecific rabbit IgG was used in control immunoprecipitations. A representative blot from three independent experiments is shown. WCL were analyzed for HIF1α and Spry2. B, HuH7 cells transfected with indicated siRNAs (20 nm each) were incubated in hypoxia for 8 h and treated with MG132 (10 μm) for the last 2 h of incubation. Immunocytochemistry was performed as described under “Experimental Procedures” to stain for Spry2 (green) and HIF1α (red). Images shown are ×126 magnification. White bar (bottom left) is the scale for 20 μm. C, HuH7 cells were incubated in hypoxia (3% O₂) for 8 h. A PLA reaction was performed as described under “Experimental Procedures” to look at Spry2 and HIF1α interaction. Images shown are ×63 magnification, and the inset (top left) is three times greater. White bar (bottom right) is scale for 100 μm. Controls included PLA secondary antibodies (secondary only) alone and complete PLA assay without HIF1α antibody (omit one primary). Graph shows mean ± variance of the quantified puncta divided by the number of cells in the field with the background from the control with 2° antibody only (no 1° antibody) subtracted and normalized to control siRNA-transfected cells for two independent experiments in which 10–15 fields were analyzed for each condition.

Previous work from our laboratory has shown that Spry2 is also hydroxylated by PHDs and can bind to pVHL (30). With this in mind and the data demonstrating the existence of endogenous Spry2 in a complex with HIF1α (Fig. 5), we postulated that Spry2 by associating with HIF1α brings more pVHL in proximity of HIF1α, and therefore, it enhances the degradation of HIF1α. To investigate the involvement of pVHL in the ability of Spry2 to regulate HIF1α, HuH7 cells were transfected with two different siRNAs targeting pVHL and the effect of Spry2 silencing on HIF1α protein was monitored. Indeed, as shown previously (Fig. 1A), silencing of Spry2 resulted in a significant increase in HIF1α protein levels without altering pVHL protein levels (Fig. 6A). However, once pVHL is silenced with either of the two pVHL targeting siRNAs, the silencing of Spry2 no longer affected HIF1α protein levels (Fig. 6A). These data suggest that Spry2 regulates HIF1α in a pVHL-dependent mechanism.

FIGURE 6.

Spry2 enhances the degradation of HIF1α in a pVHL-dependent manner. A, HuH7 were transfected with control (Cntrl) or Spry2 siRNAs along with control (−) or one of the two pVHL siRNAs (1 or 4) and cultured in hypoxia (3% O₂) for 8 h. Representative Western blots probed for HIF1α, Spry2, pVHL, and actin (loading) are shown. Graph below shows mean ± S.E. of densitometric analysis of HIF1α normalized to actin from four independent experiments. B, HEK293T cells transfected with empty vector or vector constructs to express WT-Spry2 or 3P/3A-Spry2 were cultured in hypoxia (3% O₂) for 16 h. Representative Western blots probed for HIF1α, HA-Spry2, actin, and ERK1/2 are shown. Graph below shows mean ± S.E. of densitometric analysis of HIF1α normalized to the average of actin and ERK1/2 from four independent experiments. C, control and HA-Spry2 (either WT-Spry2 or 3P/3A-Spry2) expressing HeLa cells were cultured in hypoxia for 24 h. Representative Western blots probed for HIF1α, HA-Spry2, and tubulin (loading control) are shown. Graph below shows mean ± S.E. of densitometric analysis of HIF1α normalized to tubulin from four independent experiments. D, HEK293T cells were transfected with constructs of HIF1α and FLAG-tagged ubiquitin along with constructs of WT-Spry2 or 3P/3A-Spry2. Cells were cultured in hypoxia (3% O₂) for 16 h, and MG132 (10 μm) was added in during the last 2 h of incubation. HIF1α was IP, and the amount of ubiquitylated HIF1α was monitored by Western blotting analyses; immunoprecipitation with nonspecific mouse IgG served as control. WCL were analyzed for HIF1α, Spry2, FLAG-tagged ubiquitin and tubulin (loading). Western blot images were captured using the Bio-Rad ChemiDoc XRS+ instrument. Graph shows mean ± S.E. of densitometric analysis of ubiquitylated HIF1α divided by total immunoprecipitated HIF1α from three independent experiments. E, HuH7 cells transfected with control or Spry2 siRNAs were incubated in hypoxia (3% O₂) for 8 h. pVHL was IP, and the co-immunoprecipitation of HIF1α was monitored. Western blot images were captured using the Bio-Rad ChemiDoc XRS+ instrument. Graph below shows mean ± S.E. of densitometric analysis of HIF1α normalized to pVHL immunoprecipitated from four independent experiments. Statistical significance was assessed by unpaired Student's t test (A and C–E) and one-way ANOVA (B) with Tukey's multiple comparisons; n.s., not significant.

To further investigate the contribution of pVHL in the ability of Spry2 to regulate HIF1α, we utilized a mutant form of Spry2 that cannot bind pVHL. Previously, using proteomic analyses we showed that Pro-18, Pro-144, and Pro-160 on Spry2 are hydroxylated (30). The substitution of these three Pro residues to Ala abolished the ability of Spry2 to bind pVHL (30). Here, we have designated this 3Pro → 3Ala substituted form of Spry2 as “3P/3A-Spry2.”

To determine whether the 3P/3A-Spry2 could alter HIF1α protein levels, HEK293T cells were transfected with empty vector, wild-type (WT)-Spry2, or 3P/3A-Spry2, and HIF1α levels were monitored. As expected and shown for HeLa cells (Fig. 2C), the expression of WT-Spry2 resulted in a 50% reduction in HIF1α protein levels compared with empty vector transfected cells (Fig. 6B). Intriguingly, the expression of 3P/3A-Spry2, which cannot bind pVHL, had no significant effect on HIF1α protein levels compared with vector-transfected cells (Fig. 6B). Similar results were observed in HeLa cells stably expressing no Spry2, WT-Spry2, or 3P/3A-Spry2 (Fig. 6C). Once again, HeLa cells expressing WT-Spry2 had a 50% reduction in HIF1α protein levels, although the expression of 3P/3A-Spry2 had no significant effect on HIF1α levels when compared with cells not expressing Spry2 (Fig. 6C). These data provide further evidence that Spry2 regulates HIF1α in a pVHL-dependent manner because when Spry2 cannot bind pVHL, it cannot alter HIF1α protein levels.

To determine whether the ubiquitylation of HIF1α was altered when WT-Spry2 or 3P/3A-Spry2 were overexpressed in HEK293T cells, we transfected these two forms of Spry2 together with FLAG-ubiquitin and HIF1α and monitored HIF1α ubiquitylation after immunoprecipitating HIF1α. As shown in Fig. 6D, ubiquitylation of HIF1α in cells expressing WT-Spry2 was approximately twice as high as that in cells expressing 3P/3A-Spry2.

We reasoned that if Spry2 was bringing more pVHL in proximity to HIF1α, then in immunoprecipitates of pVHL the amount of HIF1α co-immunoprecipitating should be diminished when Spry2 is silenced by siRNA. Indeed, when Spry2 was silenced in HuH7 cells, immunoprecipitates of pVHL contained ∼50% less HIF1α (Fig. 6E). Notably, the protein levels of PHD1, PHD2, PHD3, and pVHL were not significantly altered by the expression of WT-Spry2 or 3P/3A-Spry2 compared with empty vector-transfected cells (Fig. 7A).

FIGURE 7.

3P/3A-Spry2 interacts with HIF1α and inhibits phosphorylation of AKT to a similar extent as WT-Spry2. A, HEK293T lysates from Fig. 6B were used to monitor the levels of pVHL, PHD1, PHD2, PHD3, Spry2, actin, and ERK1/2 by Western blotting analyses. A representative blot is shown, and the graph shows mean ± S.E. of densitometric analysis of protein of interest normalized to the average of actin and ERK1/2 from four independent experiments. B, HEK293T cells transfected with vector constructs to express HIF1α and either WT-Spry2 or 3P/3A-Spry2 were cultured in hypoxia (3% O₂) for 16 h. HIF1α was IP, and the co-immunoprecipitation of Spry2 was monitored; immunoprecipitation with nonspecific IgG was used a control. Western blot images were captured on the Bio-Rad ChemiDoc XRS+ instrument. A representative blot from three independent experiments is shown. WCLs were probed for HIF1α, HA-Spry2, and actin (loading). C, HEK293T lysates from Fig. 6B were used to probe for phospho-AKT Ser-473, and HA-Spry2 and actin (loading) are shown. Graph below shows mean ± S.E. of densitometric analysis of pAKT normalized to actin from four independent experiments. Statistical significance was assessed by a one-way ANOVA with Tukey's multiple comparisons.

3P/3A-Spry2 Can Associate with HIF1α and Inhibit the Phosphorylation of AKT

To address the possibility that the mutation of the three proline residues in 3P/3A-Spry2 abrogated its ability to associate with HIF1α and therefore abolished its ability to alter HIF1α protein levels, we investigated whether the 3P/3A-Spry2 interacted with HIF1α. HEK293T cells were transfected with HIF1α and either WT-Spry2 or 3P/3A-Spry2 and placed in hypoxia. The WCL indicates that equal amounts of the two Spry2 constructs were transfected into the cells (Fig. 7B). Because both WT-Spry2 and 3P/3A-Spry2 co-immunoprecipitated with HIF1α (Fig. 7B), the substitution of the three prolines on Spry2 with alanines does not alter the ability of Spry2 to associate with HIF1α.

Spry2 has been shown to inhibit the downstream signaling processes such as phosphorylation and activation of AKT (22, 67, 68). Therefore, we also determined whether the 3P/3A-Spry2 retained its ability to modulate downstream signaling such as phosphorylation of AKT. Using the HEK293T cell lysate from experiments described in Fig. 6B, the levels of phosphorylation of AKT (pAKT) on Ser-473 were monitored. As expected from our previous findings (69), the expression of WT-Spry2 reduced pAKT content by more than 50%. Likewise, 3P/3A-Spry2 also reduced pAKT levels to a similar extent (Fig. 7C). Thus, 3P/3A-Spry2 retains its ability to inhibit the phosphorylation of AKT. Taken together, these data indicate that 3P/3A-Spry2 only lacks the ability to bind to pVHL (30) and to alter HIF1α levels (Fig. 6, B and C) but retains its other functions such as binding to HIF1α and inhibiting AKT phosphorylation (Fig. 7, B and C).

WT-Spry2, but Not 3P/3A-Spry2, Regulates HIF1α-sensitive Glucose Uptake

Because 3P/3A-Spry2 does not alter HIF1α protein levels, we reasoned that a biological function that is regulated by HIF1α would be susceptible to modulation by WT-Spry2 but not 3P/3A-Spry2. In this context, it is well established that HIF1α regulates the expression of certain glucose transporters, such as GLUT1 and GLUT3, and the increased expression of these glucose transporters enhances glucose uptake into cells (59, 70). Therefore, in HuH7 and HeLa cells, we monitored HIF1α-sensitive glucose uptake after transfecting either WT- or 3P/3A-Spry2. As shown in Fig. 8A, 2-deoxy-d-[³H]glucose uptake in HeLa cells exposed to hypoxia was linear over time, and silencing of HIF1α decreases 2-deoxy-d-[³H]glucose uptake by 31%. Because HIF1α does not regulate all isoforms of glucose transporters, silencing of HIF1α only diminishes that portion of the glucose transport that is HIF1α-dependent; in the case of HeLa cells, this is 30% of the total glucose uptake in hypoxia. Here, we refer to this as the HIF1α-sensitive glucose uptake. Moreover, the addition of unlabeled 2-deoxy-d-glucose markedly inhibited 2-deoxy-d-[³H]glucose uptake in the presence and absence of HIF1α demonstrating the specificity of glucose uptake via glucose transporters (Fig. 8A).

FIGURE 8.

WT-Spry2, but not 3P/3A-Spry2, reduces HIF1α-sensitive 2-deoxy-d-[³H]glucose uptake. A, HeLa cells transfected with control (Cntrl) (closed circle) or HIF1α-specific siRNA (closed square) were cultured in hypoxia for 24 h. 2-Deoxy-d-[³H]glucose (100 μm) uptake was monitored over the time course shown following the protocol described under “Experimental Procedures.” Samples treated in an identical manner but with excess unlabeled 2-deoxy-d-glucose (20 mm) were used a control to demonstrate specificity (open circle and square). Cells were lysed, and the amount of radioactivity was measured in a scintillation counter, and a protein determination was performed on the lysates. Data are presented as picomoles of 2-deoxy-d-[³H]glucose uptake per μg of protein. HeLa cells (B) and HuH7 cells (C) were transfected with empty vector or vector constructs to express WT-Spry2 or 3P/3A-Spry2 along with control (Cntrl) or HIF1α-specific siRNA. 2-Deoxy-d-[³H]glucose uptake over 10 min was monitored. HIF1α-sensitive uptake was calculated by subtracting the 2-deoxy-d-[³H]glucose uptake from HIF1α siRNA-transfected cells from the uptake in control siRNA-transfected cells. Graph shows mean ± S.E. of picomoles of HIF1α-sensitive 2-deoxy-d-[³H]glucose per μg of protein from three independent experiments. Statistical significance was assessed by one-way ANOVA with Dunnett's multiple comparisons (B) or unpaired student t test (C), n.s., not significant.

For the experimental data presented in Fig. 8, B and C, by monitoring 2-deoxy-d-[³H]glucose uptake with or without HIF1α silencing, we determined the amount of HIF1α-sensitive glucose uptake in cells transfected with vector, WT-Spry2, or 3P/3A-Spry2. In HeLa cells, the HIF1α-sensitive portion of glucose uptake was significantly inhibited (61%) by WT-Spry2, but it was not affected by 3P/3A-Spry2 (Fig. 8B). Similarly, in HuH7 cells, WT-Spry2 reduced HIF1α-sensitive 2-deoxy-d-[³H]glucose uptake by 73%, whereas 3P/3A-Spry2 did not affect it (Fig. 8C). Additionally, in HuH7 cells, total glucose uptake was elevated by ∼3.5-fold upon exposure to hypoxia, and interestingly, the transfection of WT-Spry2 or 3P/3A Spry2 did not alter glucose uptake in cells incubated in normoxia in which HIF1α protein levels would be much lower (data not shown). These findings are consistent with the observations that WT-Spry2, but not 3P/3A-Spry2, regulates HIF1α protein content in cells (Fig. 6, B and C).

Discussion

Sprouty proteins, especially Spry2, working via a number of different mechanisms, have been shown to modulate signaling downstream of receptor tyrosine kinases to inhibit cell proliferation and migration (20, 21, 71, 72). Consistent with these findings, several studies have shown that Spry2 levels are decreased in different forms of cancer, including those of the breast, liver, prostate, and lung, suggesting a potential role for Spry2 as a tumor suppressor (6–11, 17–19). Because decreased protein levels of Spry2 in hepatocellular carcinomas have been associated with poor patient survival and prognosis, it has been suggested that Spry2 levels could be utilized as a prognostic marker (73). However, the role of HIFα subunits, in particular HIF1α, is now well established in promoting tumor survival, growth, and metastasis. Specifically, HIF1α by augmenting the transcription of genes encoding for glucose transporters, GLUT1 and GLUT3, as well as several glycolytic enzymes, provides tumors with a growth advantage in hypoxia (54–60). Hence, although Spry2 may have tumor suppressor functions, HIFα subunits, especially HIF1α, act as tumor promoters (47–49). In this study, using HIF1α as a prototype of HIFα subunits, we have uncovered a novel mechanism for regulation of HIFα protein levels by Spry2. Our findings, for the first time, demonstrate that endogenous Spry2 regulates HIF1α protein levels in a variety of different cell types derived from breast (MCF-7 and MDA-MB-231), liver (HuH7 and Hep3B), and lung (A549) cancer in which Spry2 (Fig. 2B) plays a critical role. Additionally, silencing of Spry2 in HEK293T (Fig. 4B) or its ectopic expression in these cells and HeLa cells regulates HIF1α protein levels (Figs. 2C and 6, B and C). These observations establish the generality of our findings. The silencing of endogenous Spry2 elevates HIF1α and HIF2α protein levels (Fig. 1, A and C) and augments their ability to regulate their respective target genes GLUT1, PGK1, and PFK for HIF1α and EPO for HIF2α (Fig. 3, A–D). Consistently, the overexpression of Spry2 (Figs. 2C and 6B) or other Spry isoforms, Spry1, Spry3, and Spry4 (data not shown), decrease HIF1α levels demonstrating the ability of different Spry isoforms to modulate HIF1α levels. Hence, the findings described in this study are also applicable to the other Spry isoforms.

It is well established that pVHL and associated E3 ligase degrades HIF1α and HIF2α under normoxic conditions via a process that involves their hydroxylation by PHDs (41, 42). In hypoxia, the lack of free molecular oxygen reduces, but does not completely inhibit, activities of the PHDs (74, 75). This results in a reduction of pVHL/associated E3 ligase binding to HIFα subunits, and therefore, a decrease in their ubiquitylation and degradation in hypoxia (39, 40). Herein, we demonstrate that Spry2 decreases the amount of HIF1α by increasing its ubiquitylation and degradation even in hypoxic conditions suggesting that Spry2 enhances the proteosomal degradation of HIF1α (Fig. 4). This process requires pVHL because silencing of pVHL abrogated the ability of Spry2 to regulate HIF1α (Fig. 6A).

Furthermore, for the reasons described below, our data suggest that Spry2, which we have previously shown to bind pVHL (30), exists in a complex with HIF1α and thereby brings more pVHL in the proximity of HIF1α to permit its ubiquitylation and degradation. First, using three different approaches, we demonstrate that Spry2 and HIF1α exist in a complex (Fig. 5). Second, the silencing of pVHL expression abrogates the increase in HIF1α levels when Spry2 is silenced, and the expression of WT-Spry2, but not 3P/3A-Spry2 that does not bind pVHL (30), decreases HIF1α levels (Fig. 6, A–C). Notably, 3P/3A-Spry2 is as effective as WT-Spry2 in attenuating AKT phosphorylation (Fig. 7C) and associating with HIF1α (Fig. 7B). Thus, 3P/3A-Spry2 does not lose all of its biological functions. Third, ubiquitylation of HIF1α when WT-Spry2 is overexpressed is enhanced 2-fold as compared to when 3P/3A-Spry2 is overexpressed (Fig. 6D), and immunoprecipitates of pVHL from HuH7 cells in which Spry2 has been silenced contain ∼50% less HIF1α (Fig. 6E). All of this evidence strongly supports the notion that Spry2, by interacting with both pVHL and HIF1α, brings more pVHL in proximity of HIF1α and thereby enhances HIF1α ubiquitylation and degradation. The latter is supported by our data that show silencing of Spry2 diminishes the ubiquitylation of HIF1α (Fig. 4B). Consistent with our proposal that Spry2 acts as a scaffold to bring more pVHL/associated E3 ligase in proximity of HIF1α, unlike WT-Spry2, the 3P/3A-Spry2 that does not bind pVHL also does not inhibit HIF1α-sensitive glucose transport in HeLa and HuH7 cells (Fig. 8). The paradigm we propose fits well with the function of Spry2 as a scaffolding protein, which by interacting with other proteins regulates different signaling pathways and biological functions (20–25).

One of the interesting and novel observations by immunocytochemistry and PLA (Fig. 5, B and C) from our studies is the nuclear co-localization of Spry2 with HIF1α. Although we and others have shown cytoplasmic and membrane localization of Spry2 (65, 66), to date its nuclear localization has not been reported. Notably, Spry2 does not have a nuclear localization sequence, and how it enters the nucleus is not known. However, as clearly shown here, it can interact with nuclear proteins such as HIF1α and perhaps other nuclear proteins not yet identified. This adds a novel dimension by which Spry2 can regulate biological processes.

Although increasing HIF stability by inhibiting pVHL has been shown previously (76–78), only one report, aside from this study, has shown that HIF1α degradation by pVHL can be enhanced in hypoxia. Specifically, that report showed that SSAT2 binding to pVHL stabilized the interactions between pVHL and elongin C and enhanced the ability of pVHL to degrade HIF1α (79). Likewise, it has been shown that recruitment of PHDs in the proximity of HIF1α can enhance its degradation in hypoxia. Osteosarcoma-amplified 9 (OS-9) and mitogen-activated protein kinase organizer 1 (MORG1) by associating with PHDs and bringing them in proximity to HIF1α enhance the hydroxylation and therefore the degradation of HIF1α in hypoxia (80, 81). Furthermore, by a mechanism similar to what we have shown here, HIF1α degradation is enhanced by p53 interacting with HIF1α and bringing the E3 ubiquitin ligase MDM2 in proximity of HIF1α (82). Altogether, these previous reports support the concept that a protein can enhance the degradation of another protein by bringing it into proximity to an enzyme that causes its degradation. However, our findings reported here for the first time demonstrate that Spry2 by scaffolding HIF1α and pVHL can augment the ubiquitylation of HIF1α and decrease its stability.

Reports in the literature suggest that AKT and ERK can regulate HIF1α translation, protein levels, and activity (83–88). The role of AKT regulation of HIF1α appears to be cell type-specific and not critical for HIF1α stabilization in hypoxia (89). Nevertheless, the ability of Spry2 to inhibit ERK and AKT activation could provide a possible mechanism by which Spry2 regulates HIF1α protein levels. However, we have shown that both WT-Spry2 and 3P/3A-Spry2 inhibit the phosphorylation of AKT (Fig. 7C), but only WT-Spry2 decreases HIF1α protein levels suggesting AKT activity has no role in the ability of Spry2 to regulate HIF1α (Fig. 6, B and C). Likewise, under the same conditions as those in Fig. 1, silencing of Spry2 in HuH7 cells did not alter ERK1/2 phosphorylation status suggesting that ERK1/2 do not contribute to the elevation of HIF1α protein levels (data not shown).

One study by Liu et al. (84) demonstrated that microRNA-21 (miR-21) elevated HIF1α levels due to modest increases in AKT and ERK1/2 activation. However, miR-21 also decreases the expression of Spry2 (90). Hence, in light of our findings, one could propose that in the studies of Liu et al. (84) miR21-mediated decrease in Spry2 levels, which was not monitored in that study, contributes to the elevation in HIF1α protein levels.

Our findings that silencing of endogenous Spry2 elevates the transcription of HIF1α-responsive genes (Fig. 3, A–C) and that Spry2 also inhibits HIF1α-sensitive glucose uptake (Fig. 8) implicate Spry2 as a regulator of glycolysis. In this context, Wang et al. (91) showed that expression of a dominant negative Spry2 with AKT resulted in increased membrane levels of GLUT1 and GLUT4 as well as an increase in the glycolysis master regulator pyruvate kinase M2 (PKM2). However, that study did not delineate the mechanism by which Spry2 up-regulated PKM2. Given our findings in this report and the fact that PKM2 transcription and expression are elevated by HIF1α (92), it is tempting to speculate that the up-regulation of PKM2, GLUT1, and GLUT4 by dominant negative Spry2 is due to an elevation of HIF1α levels because dominant negative Spry2 would ablate the ability of endogenous Spry2 to decrease HIF1α levels.

In summary, we showed that Spry2 acts as a scaffold that complexes with HIF1α and pVHL elevating the local concentrations of pVHL to ubiquitylate HIF1α in hypoxia and target it for degradation. The regulation of HIF1α levels by Spry2 also translates into regulation of HIF1α-responsive genes and glucose uptake. Thus, Spry2 also inhibits the expression of glycolytic genes and HIF1α-sensitive glucose uptake. Because glucose uptake and glycolysis are critical in providing survival advantage and promotion of cancer, our data suggest Spry2, in part, may exert its “tumor suppressor” actions by regulating HIF1α protein levels and gene transcription. As a corollary, when Spry2 is decreased or lost in certain forms of cancer, the transcription of HIF1α-regulated genes would be elevated and further promote the progression of the cancer. Overall, these findings reveal a novel mode of action for Spry2 in regulating cellular signaling and biological functions such as glucose uptake.

Experimental Procedures

Chemicals and Reagents

Cycloheximide and N-ethylmaleimide were purchased from Sigma. MG132 was purchased from Selleck Chem (Houston, TX). All primers, probes, and siRNAs were synthesized by Integrated DNA Technologies, Inc. (Coralville, IA). The sequences of the various siRNAs used are described in supplemental Table 1A. Antibodies used for different applications and their dilutions are listed in supplemental Table 1B.

DNA Constructs

The cloning of human Spry2 cDNA into the pHM6-HA-vector is described elsewhere (93). pHM6-HA-3P/3A-Spry2 was generated using site-directed mutagenesis of the wild-type pHM6-HA-Spry2 to mutate prolines 18, 144, and 160 (30). The cDNA of HIF1α was purchased from Origene and subcloned into pcDNA3 vector using HindIII and NotI restriction sites. pRG-TK was kindly provided by Dr. Neil Clipstone, Loyola University Chicago. pGL2-Pfkfb3/−3566 was kindly provided by Ramon Bartrons, University of Barcelona. pCMV-3×-FLAG-ubiquitin was kindly provided by Dr. Adriano Marchese, Medical College of Wisconsin. The cloning of FLAG-Spry2 was described previously (29). FLAG-Spry1, FLAG-Spry3, and FLAG-Spry4 were generous gifts from Dr. Graeme Guy, Institute of Molecular and Cell Biology, Singapore. All constructs were verified by sequencing.

Cell Culture

HuH7 and Hep3B cells were kindly provided by Dr. Basabi Rana, University of Illinois at Chicago. HuH7 cells were cultured in DMEM/nutrient mixture F-12 50:50 supplemented with HEPES (10 mm), 10% FBS, penicillin (100 units/ml), and streptomycin (100 μg/ml). Hep3B cells were cultured in MEM supplemented with HEPES (10 mm), sodium pyruvate (1 mm), MEM non-essential amino acid solution (1×), 10% FBS, penicillin (100 units/ml), and streptomycin (100 μg/ml). HEK293T cells were kindly provided by Dr. Jody Martin, Loyola University, Chicago. HeLa cells were kindly gifted by the late Dr. Jill Lahti, St. Jude Children's Research Hospital, Memphis, TN. HEK293T and HeLa cells were cultured in high glucose containing DMEM supplemented with 10% FBS, penicillin (100 units/ml), and streptomycin (100 μg/ml). MDA-MB-231 and MCF7 cells were kindly provided by Dr. Ajay Rana, University of Illinois at Chicago or Dr. Paula McKeown-Longo, Albany Medical College. MDA-MB-231 cells were cultured in RMPI 1640 medium or DMEM supplemented with 10% FBS, penicillin (100 units/ml), and streptomycin (100 μg/ml). MCF7 cells were cultured in MEM supplemented with 0.01 mg/ml human recombinant insulin, 10% FBS, penicillin (100 units/ml), and streptomycin (100 μg/ml) or DMEM as described previously. A549 cells were kindly provided by Dr. Maurizio Bocchetta, Loyola University Chicago, or Dr. Paula McKeown-Longo, Albany Medical College, and were cultured in F-12K supplemented with 10% FBS, penicillin (100 units/ml), and streptomycin (100 μg/ml). Cells were maintained under normoxic conditions (21% O₂) in a water-jacket CO₂ incubator (Thermo Fisher Scientific, Waltham, MA) at 37 °C, 5% CO₂. Cells were maintained under hypoxic conditions (3% O₂) in a Coy Hypoxic Chamber (Grass Lake, MI) at 37 °C, 5% CO₂. Media used for hypoxic experiments were pre-equilibrated under hypoxia for 16 h.

Silencing of Endogenous Spry2

Cells (150,000–250,000/35-mm dish) were plated, and the next day they were transfected using Trans-IT TKO (Mirus Bio LLC, Madison, WI) transfection reagent with 20 nm 27-mer (control) mutant Spry2 siRNA containing three ribonucleotide substitutions of the Spry2 siRNA or Spry2 siRNA duplex following the manufacturer's protocol. The cells were placed in hypoxia the next day, and the media were changed to pre-equilibrated hypoxic media. Cell lysates were collected after either 8 h (for HIF1α studies) or 24 h (for HIF2α studies).

Isolating RNA, cDNA Synthesis, and qRT-PCR

HuH7 cells were plated (250,000/35-mm dish or 80,000/12-well plate) and transfected the following day with control (40 nm), Spry2 (20 nm +20 nm control (for HIF1/2α)), Spry2 (20 nm) + HIF1α (20 nm), or Spry2 (20 nm) + HIF2α (20 nm) siRNAs. After 16 or 24 h in hypoxia, the cells were lysed in TRIzol (Life Technologies, Inc.), and RNA was extracted according to the manufacturer's protocol with the addition of GlycoBlue (Life Technologies, Inc.) to aid in precipitation of RNA. At least duplicate samples were collected for each condition. The RNA was re-precipitated with sodium acetate and ethanol to remove impurities. All RNA used had a 260:280 ratio above 1.8 and a 260:230 ratio above 1.9. Total RNA (1 μg or 500 ng) was reverse-transcribed into cDNA using the SuperScript® VILO cDNA synthesis kit (Life Technologies, Inc.) following the manufacturer's protocol. The cDNA (10 ng) was used for qRT-PCR using either FastStart Universal Probe Master Mix with Rox (Roche Applied Science) and primer probe mixes or FastStart Universal SYBR Master Mix with Rox (Roche Applied Science) with primers. All primers and probes with their calculated efficiencies are listed in supplemental Table 2, A and B. The cycle threshold (CT) was calculated using a single threshold. Three housekeeping genes were used: 18S ribosomal RNA (RN18S1), hypoxanthine-guanine phosphoribosyltransferase (HPRT), and ribosomal protein, large, P0 (RPLP0). The average was taken for the CT values of the control siRNA-transfected cells for all housekeeping genes and genes of interest (GOI). The fold change of each housekeeping gene and GOI was calculated by taking the efficiency (E) of the gene to the power of the difference of the experimental CT and the average CT of control CT (fold change = E∧(CT_exp − average(CT_control)). The geometric mean was taken of three housekeeping genes. To calculate the relative mRNA amount of each GOI, the fold change of the GOI was divided by the geometric mean of the housekeeping genes (relative mRNA amount = fold change(GOI)/geomean(fold change(RN18S1, HPRT, RPLP0)).

Luciferase Assays

HuH7 cells were plated (50,000/24-well plate), and the following day were transfected using Trans-IT TKO (Mirus Bio LLC, Madison, WI) with control siRNA or siRNAs against Spry2, Spry2 + HIF1α, or Spry2 + HIF2α (as described above under “Isolating RNA, cDNA Synthesis, and qRT-PCR”) and 1 μg of pGL2-Pfkfb3/−3566 plus 10 ng of pRG-TK and incubated in normoxia or hypoxia (3% O₂) for 24 h. The cells were lysed in Passive Lysis Buffer (Promega, Madison, WI). Luciferase assays were performed using the Promega Dual-Luciferase kit and read on a PHERAStar FS plate reader (BMG Labtech, Cary, NC).

Stability of HIF1α

HuH7 cells (250,000/35-mm dish) were plated and the following day transfected with 20 nm each of control or Spry2 siRNAs as stated above. The cells were placed in hypoxia, and 8 h later cycloheximide (200 μm) was added. The cells were lysed after 0, 5, or 10 min in Laemmli sample buffer. HIF1α, Spry2, ERK1/2, and actin levels were monitored using Western blotting.

Ubiquitylation Assays

HEK293T cells (300,000/60-mm dish) were plated and the following day transfected with 1.25 μg of pJX40-FLAG-ubiquitin and 625 ng of pcDNA3-HIF1α along with either control or Spry2 siRNAs or WT-Spry2 or 3P/3A-Spry2. The following day the cells were placed in hypoxia for 12 h, and 25 μm MG132 was added for an additional 4 h for silencing of Spry2 experiments or 2 h for overexpression experiments. The cells were lysed in denaturing lysis buffer (50 mm Tris-HCl, 150 mm NaCl, 1 mm EDTA, 1 mm EGTA, 1% SDS, 1% Nonidet P-40, and 0.5% sodium deoxycholate with the following protease inhibitors: 50 mm NaF, 5 mm β-glycerol phosphate, 1 mm sodium orthovanadate, 150 nm aprotinin, 1.5 mm pepstatin A, 25 μm MG132, 4.7 mm leupeptin, 6.6 mm benzamidine, 8.3 mm sodium pyrophosphate, 5 mm N-ethylmaleimide, 0.1 mm phenylmethylsulfonyl fluoride, pH 7.4). The samples were sonicated twice for 10 s on ice. Lysate (100 μg) was diluted 1:10 with lysis buffer without SDS and incubated with 3 μg of anti-FLAG-M2, anti-HIF1α, or mouse IgG for 1 h. Protein G beads (Roche Applied Science) (20 μl slurry) were incubated with the lysate for 2 h. The beads were washed three times with lysis buffer with 0.1% SDS. Heating in Laemmli sample buffer eluted the bound proteins, which were then separated by SDS-PAGE, and HIF1α and FLAG-ubiquitin were detected by Western blotting analysis.

Co-immunoprecipitation (Co-IP) of Endogenous Spry2 and HIF1α

HuH7 cells (800,000/60-mm dish) were plated and the following day exposed to hypoxia for 8 h with the proteosomal inhibitor MG132 (25 μm) treatment during the last 4 h. Cells were washed twice with ice-cold PBS, lysed in lysis buffer (as described above except with 1% Triton X-100 instead of SDS, Nonidet P-40, and sodium deoxycholate), and rotated for 30 min at 4 °C. Lysates were centrifuged at 15,000 × g for 15 min, and 500 μg of protein was rotated with 3 μg of anti-N-terminal Spry2 antibody overnight at 4 °C. The next day the lysates were incubated rotating with 30 μl of a slurry of protein G beads (Roche Applied Science) for 2 h at 4 °C. The beads were washed three times in lysis buffer and then heated to 95 °C for 10 min in Laemmli buffer to collect bound proteins, which were analyzed for HIF1α and Spry2 by Western blotting.

ICC and PLA

HuH7 cells were plated in an 8-well chamber slide and the following day transfected with 20 nm each of control (for Spry2) or Spry2 siRNAs (ICC only) or control (for HIF1α) or HIF1α siRNAs (ICC and PLA) as stated above. The next day cells were placed in hypoxia for 8 h and then fixed with 3.7% formaldehyde in PIPES, pH 6.8, for 10 min at room temperature, washed, permeabilized with 0.1% Triton X-100 for 10 min, and blocked with 10% normal goat serum for 1 h at room temperature (ICC) or the PLA kit (Sigma) blocking solution for 30 min at 37 °C (PLA). Rabbit Spry2 (Rockland, Limerick, PA) (1:400 dilution) and mouse HIF1α (BD Biosciences) (1:50 dilution) were incubated overnight 4 °C. For ICC, the secondary antibodies of goat anti-rabbit conjugated with Alexa Fluor 488 or goat anti-mouse conjugated to Alexa Fluor 594 (ThermoFisher, Grand Island, NY) (1:500 dilution) were incubated on the slide for 1 h. For PLA (Sigma), the manufacturer's protocol was followed. Confocal images were obtained with a TCS SP5 laser scanning confocal microscope (Leica, Exton, PA) equipped with DMI6000 inverted microscope with blue diode (405 nm), argon (458/476/488/496/514 nm), diode pumped solid state (561 nm), and HeNe (633 nm) lasers and a ×63 HCX PL APO λblue (numerical aperture, 1.4 oil immersion) objective lens. The software used to capture the images was Leica Application Suite Advanced Fluorescence. For ICC, five fields were taken for each condition. For PLA, 10–15 fields were taken of each condition, and PLA puncta were quantified using ImageJ64 software (National Institutes of Health) using the “analyze particles” function. The ×3 zoomed insets were generated using “Zoom in images and stacks” macro by Gilles Carpentier.

Co-IP of HIF1α with pVHL

HuH7 cells (650,000/60-mm dish) were plated and the next day transfected with 20 nm control or Spry2 siRNAs as stated above. The following day, the cells were exposed to hypoxia for 8 h with proteosomal inhibitor MG132 (10 μm) treatment during the last 4 h. Immunoprecipitation of pVHL was performed using the buffer described under “Ubiquitylation Assays” except the buffer contained 0.1% SDS instead of 1% SDS. The samples were centrifuged at 15,000 rpm for 15 min, and 250 μg of protein was rotated with 3 μg of anti-pVHL (BD Biosciences)-bound bead slurry (15 μl/IP) (Roche Applied Science) overnight at 4 °C. The next day the beads were washed three times with the buffer diluted 1:1 with PBS. The bound proteins were eluted in Laemmli buffer and analyzed using Western blotting for HIF1α and pVHL.

Co-IP of WT-Spry2 and 3P/3A Spry2 with HIF1α

HEK293T cells (500,000/60-mm dish) were plated, and the next day the cells were transfected with 750 ng of pHM6-vector, pHM6-WT-HA-Spry2, or pHM6–3P/3A-HA-Spry2 along with 750 ng of pcDNA3-HIF1α with TransIT-2020 (Mirus Bio LLC, Madison, WI) following the manufacturer's protocol. The next day the cells were incubated in hypoxia for 16 h. Immunoprecipitation protocol was modified from the two-step lysis method in Ref. 94. Protein G bead slurry (Roche Applied Science) was incubated overnight at 4 °C with 2 μg of antibody (anti-HIF1α, Novus Biologicals, Littleton, CO) or mouse IgG (Sigma) in PBS with 0.5% CHAPS and 5% BSA. The next day the cells were lysed in 500 μl of Triton X-100 hypotonic buffer (94) with the addition of protease and phosphatase inhibitors stated above. The samples were sonicated twice at 15% power for 10 s. NaCl was added to each tube to a final concentration of 420 mm, and the samples were incubated on ice for 1 h. The samples were sonicated again as stated previously and centrifuged at 10,000 × g for 10 min. The supernatant (150 μg) was incubated with antibody-bound beads (either mouse IgG or HIF1α) for 2 h. The samples were washed three times in the hypotonic buffer with 150 mm NaCl. The final wash was removed, and the bound proteins were eluted off the beads by heating in Laemmli sample buffer.

pVHL Silencing

HuH7 cells were plated (250,000/35-mm dish) and transfected with 20 nm each of control or Spry2 siRNAs along with 40 nm each of control or one of two pVHL siRNAs (pVHL siRNA1 or siRNA4). The next day the cells were incubated in hypoxia for 8 h. The cells were lysed in Laemmli sample buffer, and the levels of HIF1α, Spry2, actin, and pVHL were monitored using Western blotting.

Expression of Wild-type (WT) Spry2 and Spry2 P18A/P144A/P160A (3P/3A) Mutant Spry2

HEK293T cells were plated (250,000/35-mm dish) and transfected the next day using TransIT-2020 transfection reagent (Mirus Bio LLC, Madison WI) with 250 ng of pHM6 vector, pHM6-WT-HA-Spry2, or pHM6–3P/3A-HA-Spry2 following the manufacturer's protocol. After 24 h, the cells were placed in hypoxia for 16 h. The cells were lysed in Laemmli sample buffer and analyzed for HIF1α, HA-Spry2, actin, ERK1/2, pAKT Ser-473, PHD1, PHD2, PHD3, and pVHL levels using Western blotting.

Glucose Uptake Assays

HeLa or HuH7 cells (40,000/well of 24-well plate) were plated and the following day transfected with 20 nm each of control or HIF1α siRNAs and 200 ng each of either pHM6-vector, pHM6-WT-HA-Spry2, or pHM6–3P/3A-HA-Spry2 using TransIT-HeLaMONSTER (Mirus) (HeLa) or TransIT-X2 (Mirus) (HuH7). Quadruplicate sets of wells for each condition were transfected. The next day the cells were placed in the hypoxic chamber, and the media were changed to pre-equilibrated hypoxic media. After 24 h, the cells were washed with Krebs-Ringer HEPES Buffer (KRH) (modified from Cold Spring Harbor Protocols) and then incubated in KRH with 100 μm 2-deoxy-d-[³H]glucose (0.5 μCi/well) for 10 min. One set of duplicate wells was incubated with an additional 20 mm 2-deoxy-d-glucose to compete with 2-deoxy-d-[³H]glucose and determine specific glucose uptake. The cells were then washed three times in ice-cold KRH and lysed in 0.5 m NaOH with 0.1% Triton X-100. An aliquot (10 μl) of the lysate from each sample was used for protein determination. Another aliquot of the lysate (140 μl) was transferred to a scintillation vial. Each well was washed once with 300 μl of KRH that was also transferred to the scintillation vials for counting after mixing with 5 ml of scintillation fluid. The vials were counted on a Beckman Coulter scintillation counter. Data were calculated to represent picomoles of 2-deoxy-d-[³H]glucose uptake/μg of total protein. To determine the HIF1α-specific glucose uptake, the average of the HIF1α siRNA-transfected samples was subtracted from the control siRNA-transfected samples.

Statistical Analysis

All data are represented as means ± S.E. of the mean. Data were analyzed as indicated in the figure legends with two-tailed unpaired t tests, one-way analysis of variance (ANOVA), or two-way ANOVA; p < 0.05 was considered significant, n.s. indicates not significant. All statistical tests were calculated using Prism 6 (GraphPad, La Jolla, CA).

Author Contributions

K. C. H. planned and designed the experiments, collected and interpreted data, and helped write the manuscript. T. B. P. planned and designed the experiments, interpreted data, helped write the manuscript.